1-s2.0-s1352231006004377-aula3-mestrado

7/28/2019 1-s2.0-S1352231006004377-aula3-mestrado

1/12

Atmospheric Environment 40 (2006) 54645475

Examination of pollution trends in Santiago de Chile with

cluster analysis of PM10 and Ozone data

E. Gramscha,, F. Cereceda-Balicb, P. Oyolac, D. von Baerd

aPhysics Department, Universidad de Santiago, Avda. Ecuador 3493, Santiago, ChilebChemistry Department, Universidad Tecnica Federico Santa Mar a, Avda. Espana 1680, Valpara so, Chile

cComision Nacional del Medio Ambiente, Valent n Letelier 13, Santiago, ChiledPharmacy Faculty, Universidad de Concepcion, Casilla 160C, Concepcion, Chile

Received 17 August 2005; accepted 25 March 2006

Abstract

Because of the high levels of pollution that Santiago de Chile experiences every year in winter, the government has set up

an air quality monitoring network. Information from this network is employed to alert people about the quality of

air and to enforce several control strategies in order to limit pollution levels. The monitoring network has 8 stations

that measure PM10, carbon monoxide (CO), sulphur dioxide (SO2), ozone (O3) and meteorological parameters.

Some stations also measure nitrogen mono- and dioxide (NOx), fine particles (PM2.5) and carbon. In this study

we have examined the PM10 and O3 data generated by this network in the year 2000 in order to determine the

seasonal trends and spatial distribution of these pollutants over a years period. The results show that concentration

levels vary with the season, with PM10 being higher in winter and O3 in summer. All but one station, show a peak in PM10at 8:00 indicating that during the rush hour there is a strong influence from traffic, however, this influence is not seen

during the rest of the day. In winter, the PM10 maximum occurs at 24:00 h in all stations but Las Condes. This maximum is

related to decreased wind speed and lower altitude of the inversion layer. The fact that Las Condes station is at a higher

altitude than the others and it does not show the PM10 increase at night, suggest that the height of the inversion layer

occurs at lower altitude. Cluster analysis was applied to the PM10 and O3 data, and the results indicate that the city has

four large sectors with similar pollution behavior. The fact that both pollutants have similar distribution is a strong

indication that the concentration levels are primarily determined by the topographical and meteorological characteristics

of the area and that pollution generated over the city is redistributed in four large areas that have similar meteorological

and topographical conditions.

r 2006 Published by Elsevier Ltd.

Keywords: Particulate matter; Ozone; Cluster analysis

1. Introduction

The high levels of pollution that are observed in

many large cities of the world have well documented

consequences for human health (Lee et al., 2000;

Dockery et al., 1997). Santiago has high levels of

ARTICLE IN PRESS

www.elsevier.com/locate/atmosenv

1352-2310/$- see front matterr 2006 Published by Elsevier Ltd.

doi:10.1016/j.atmosenv.2006.03.062

Corresponding author. Tel.: +56 2 776 80 12;

fax: +5627769596.

E-mail address: [email protected] (E. Gramsch).
http://www.elsevier.com/locate/atmosenvhttp://localhost/var/www/apps/conversion/tmp/scratch_5/dx.doi.org/10.1016/j.atmosenv.2006.03.062mailto:[email protected]:[email protected]://localhost/var/www/apps/conversion/tmp/scratch_5/dx.doi.org/10.1016/j.atmosenv.2006.03.062http://www.elsevier.com/locate/atmosenv

7/28/2019 1-s2.0-S1352231006004377-aula3-mestrado

2/12

pollution throughout most of the year, with high

PM10 levels in winter, and high O3 levels in summer.

It is common to observe an increase in the number

of childrens hospitalizations due to respiratory

diseases following a pollution event in winter to

(Ostro et al., 1999; Sanhueza et al., 1999), even anincrease in daily mortality was observed (Cifuentes

et al., 2000; Ilabaca et al., 1999). Particle mass

concentration (PM10) averages near 300 mg m3 are

frequent in the western part of Santiago (Pudahuel,

Cerrillos). Some isolated events of 500mm m3 or

more occur several times during winter (Jorquera

et al., 1998; Perez et al., 2000; Artaxo et al., 1999).

Another effect that contributes to the high particle

levels observed in winter is the temperature inver-

sion. The height of the inversion layer during winter

could be as low as 300 m at night and early morning

(Gramsch et al., 2000), in summer, it reaches20003000 m (Rutland and Garreaud, 1995). Ozone

is a secondary pollutant and its concentration

depends on the concentration of primary pollutants

(NOx, VOC) and the intensity of solar UV radia-

tion, thus ozone concentration is high during

summer. Ozone hourly maxima reach concentra-

tions levels between 100 and 150mg m3 with some

isolated events as high as 320 mg m3 in the eastern

part of the city (Las Condes) which is located

downwind from the center of the city. The Chilean

norm for ozone is 160mg m3

hourly maximum;however this norm is exceeded more than 140 days

per year.

Because of the potential health effects associated

with elevated levels of PM10 and Ozone, the

government developed a number of control strate-

gies to help reduce pollution. In 1994, the Envir-

onmental-Base Law (Conama, 1994) was passed

and directed the National Commission for the

Environment (Conama) to develop a pollution-

control plan for Santiago and its surroundings. This

plan,which was completed on July 1, 1997,

provided the framework for the decontamination

effort in Santiago. The Plan established specific

emission reduction targets for the most commonpollutants such as particulate matter with aero-

dynamic diameter o10mm (PM10), ozone (O3),

nitrogen oxides (NOx), sulphur dioxide (SO2) and

carbon monoxide (CO). The Plan also provides the

legal framework to enforce the control strategies

needed for the pollution reduction efforts in

Santiago (Conama, 1997). The first strategies

implemented in the early 1990 were directed

towards removing fixed sources like diesel electric

generators, waste burning, and large wood and coal

heaters. Afterwards, the quality of the public

transportation buses was improved, all new carswere required to have catalytic converter, and many

streets were paved. Currently, the efforts are

directed towards improving the public transporta-

tion and reducing the number of private cars used in

the city. However, nothing has been done to reduce

pollution from kerosene and wood burning for

house heating.

An important part of the plan was to set up a

network of eight monitoring stations (Macam

network) distributed around the city and operated

by the Ministry of Health. In 1995, five monitoringstations were located near the center of the city.

Later it was determined that this arrangement did

not cover areas with high pollutant levels and most

contamination events (days with high average PM10levels) were not detected. In 1997, new stations were

added and some were closed. The new configuration

has eight stations distributed across the city that

measure PM10, O3 and CO on an hourly basis.

ARTICLE IN PRESS

Table 1

Years of operation, pollutants measured and altitude over sea level for the monitoring stations of the Macam network

Station Las Condes Providencia La Paz Parque OHiggins La Florida Pudahuel El Bosque Cerrillos

First year of operation 1988 1988 1988 1988 1997 1997 1997 1997

CO X X X X X X X X

SO2 X X X X X X X X

O3 X X X X X X X X

NOX/NO2 X Until 1996 Until 1996 X X

PM10 X X X X X X X X

PM2.5 X X X X

Height (m) 700 550 530 500 500 480 470 470

X, contaminant being measured, , not measured.

E. Gramsch et al. / Atmospheric Environment 40 (2006) 54645475 5465

7/28/2019 1-s2.0-S1352231006004377-aula3-mestrado

3/12

In addition, there are several stations that also

monitor organic and elemental carbon, PM2.5, NOxand nitrate. Information on the years of operation

and pollutants measured for all the stations of the

Macam network is shown in Table 1.

Although the stations are better distributedacross the city (Schmitz, 2004; Silva and Quiroz,

2003), the new monitoring network is still not

optimized. It is for example not known whether all

sectors with high pollution levels are monitored or if

there are too many stations covering a sector with

similar concentrations levels.

The objective of this study was to perform an

analysis of the data generated by the Macam

network to determine the pollution trends in the

city and to determine which areas of the city have

similar pollution behavior and may be covered by

too many stations of the network.

2. Methods

2.1. Experimental

PM10 was measured with a Tapered Element

Oscillating Microbalance (TEOM 1400) monitors

from Rupprecht & Patachnick, Albany, New York.

The instrument uses an oscillating hollow tube with

the free end attached to a filter element. Due toaccumulation of particles, the filter mass changes

and the oscillating frequency changes, providing a

measurement of the mass. The tapered tube, filter,

and sampled air are kept at 50 1C. The sampling

interval was set to 15 min.

Ozone was measured with 400 E UV absorption

analyzers from Teledyne Instruments, Los Angeles,

CA. A 254 nm UV light signal is passed through the

sample cell where it is absorbed in proportion to the

amount of ozone present. Using the LambertBeer

law, it is possible to obtain the ozone concentration

with a lower detectable limit is of 0.6 ppb. Meteor-

ological parameters (wind speed and direction,

temperature, humidity and pressure) are measured

at all stations of the Macam network with standard

equipment with 5 min time intervals at 3 m height.

Every day, PM10 and O3 data were obtained in

the year 2000 at all eight stations of the Macam

network. Monthly and hourly averages were calcu-

lated for both pollutants. The hourly average is

obtained by selecting all the data collected at a

specific hour, and averaging over all days of the

month.

2.2. Sampling sites

The study was performed in Santiago de Chile, a

city with a population of almost 6 million. Santiago

is located in a relatively flat valley at an altitude of

500 m. There are two hills inside the city, SanCristobal, with an altitude of 800 m above sea level

and Cerro Renca of 700 m height. The Andes

mountain range is located to the east, with hills up

to 5500 m. A smaller coastal mountain range is

located in the west, with hills up to 2000 m. The map

in Fig. 1 shows the locations of the Macam network

monitoring stations and the topography of the city.

The station of the Macam network that measures

the quality of air in downtown Santiago is Parque

OHiggins. It is placed in a large park about 2 km

south of the city center and 1 km west of a major

highway with a traffic of about 60 000 vehicles perday. The area has a mixture of houses, retail and

light industries (machine shops, auto repair shops,

furniture manufacturing shops, etc.). This station

monitors PM10, PM2.5, O3, CO, SO2, elemental and

organic carbon and meteorological parameters. A

list of the monitors and the height above sea level

for all stations is given in Table 1. Two other

stations are near downtown, Providencia and La

Paz.

Providencia is a station located about 3 km east

of the city center, about 30 m north of Providenciastreet with a traffic of 40 000 vehicles per day. This

site has some commercial activity, and many office

buildings. The station is located in a park near the

Mapocho river and it is surrounded by trees. La Paz

is located in the northern part of Santiago, in

between two large roads with about 25 000 vehicles

per day that run in the northsouth direction. These

roads have a lot of commercial activity with many

small retail stores, some light industries, and a large

hospital nearby.

The stations that are located in the western part

of the city are Pudahuel and Cerrillos, primarily in

residential areas. Pudahuel station is located in the

western part of Santiago; it is placed in a small park,

near a medical clinic. Two major roads are in this

area: one towards the south with traffic of about

20 000 vehicles per day and one in the west with

about 15 000 vehicles per day. These roads show a

lot of commercial activity with many small retail

stores. The rest of the area is mainly residential.

Cerrillos is also located in the western part of

Santiago near a street with 30 000 vehicles per day.

The area has some of commercial activity with

ARTICLE IN PRESS

E. Gramsch et al. / Atmospheric Environment 40 (2006) 546454755466

7/28/2019 1-s2.0-S1352231006004377-aula3-mestrado

4/12

many light industries around. An airport is located

towards the south of the station.

Las Condes is located in the eastern part of

Santiago at an altitude of 700 m above sea level. It is

placed in a park, south of a street with about 15 000

vehicles per day. The area is primarily residential,

with some retail stores located around the larger

streets.

La Florida and El Bosque are located in the

southern part of Santiago. Both stations are located

close to large streets, in an area with growth in real

estate. La Florida station is located about 500m

north of a road with a traffic 30 000 vehicles per day,

east of another road with 55 000 vehicles and west of

a third road with about 35 000 vehicles per day. The

area has a lot of commercial activity, heavy traffic,

several residential buildings and a residential area

with one story houses. El Bosque is a station located

in the south-west part of Santiago, near a highway

with about 60 000 vehicles per day. The area has some

commercial activity with light industry but mainly

contains residential buildings and one story houses.

2.3. Traffic information

The flux of vehicles per day was obtained from

the Demand Equilibrium Model for Multimodal

Urban Transportation Networks with Multiple

User Classes (ESTRAUS). This model simulates

the operation of a citys urban transportation

system and it is used by the Ministry of Transporta-

tion to evaluate the impacts on the urban transpor-

tation system of implementing different road

infrastructure projects (highways, metro lines, bus

corridors, etc.) as well as transport policies (road

pricing, transit integrated pricing systems, street

reversibility, increase in gasoline taxes, etc.).

2.4. Statistical methods

The study was performed using cluster analysis

with the Statistical Analysis System, SAS program

V.6.12 (SAS Institute Inc.) to classify the stations

according to the distance between them. The

distance between two stations was defined with the

ARTICLE IN PRESS

Fig. 1. City of Santiago de Chile showing the location of the monitoring stations.


7/28/2019 1-s2.0-S1352231006004377-aula3-mestrado

5/12

Pearson correlation function. When the correlation

approaches one, it indicates that the temporal

behavior of the data is similar. For example, if

two stations show an increase in PM10 at rush hour

and a decrease in the afternoon, the correlation will

be close to one, independent of the concentrationlevel. Hence, it is possible to have stations with

different average pollution levels and similar tem-

poral behavior that have close correlation.

The cluster analysis procedure is realized by

requiring that the intra-variance within a group of

stations be less than a certain number, Ri. This

number is the sum of the intra-variance of the

groups divided by the total variance. Because of a

normalization procedure for the data, the total

variance of the groups is 8, because there are eight

stations. This number (Ri) determines how close the

elements of a group are to each other. The variancesare defined by

Total variance

Pni1xi X

2

N

Intravariance

PNhi1xi xh

2

N,

where xi is the hourly average for the station, xh is

the average inside the group h, X is average of all

stations, and N is the total number of elements.

A study of the data from Santiagos monitoringnetwork was done using an index of multivariate

effectiveness (Silva and Quiroz, 2003), based on

Shannon information index. They found that data

(CO, PM10, O3 and SO2) from one of the stations

(Parque OHiggins) could be reproduced by using

information from the other stations.

3. Results

3.1. Seasonal variation of PM10 data

Concentrations of particulate material (PM10 and

PM2.5) show a seasonal trend, with the highest

concentrations during winter and the lowest con-

centrations during summer in the whole city. The

average monthly PM10 concentration can be seen in

Fig. 2, showing higher concentration in March

through August. For clarity, only four stations are

shown in Fig. 2: some are located in the eastern part

of the city (Las Condes), south (La Florida), one in

the center (Parque OHiggins) and western part of

the city (Pudahuel).

The station that has the highest PM10 concentra-tions throughout the whole year is Pudahuel

(slightly higher than La Florida). In the first part

of the year (JanuaryMay) and in November and

December, Pudahuel has higher concentrations than

Parque OHiggins and Las Condes. In August

through October, Pudahuel and Parque have similar

PM10 concentration. PM10 during June was lower

because it was a rainy month (334.2 mm compared

to 10.4 mm in May and 40.8 mm in July), indicating

that when the ground is wet less large particles are

re-suspended and PM10 is washed out from theatmosphere by rain. In Santiago, the winter months

(MayAugust) are cold with moderate rain and low

wind speeds. Summer is hot and dry and the average

wind speed is higher than the other months. An

analysis of the wind pattern can help explain the

PM10 trends.

The wind pattern in Santiago is complicated

because of the complex topography. The city is

surrounded by two mountain ranges with several

isolated hills in between. In the afternoon, in the

eastern part of the city the wind is from west to east,

from the valley towards the mountains and, at

night, the direction reverses to an east to west

direction (from the mountains towards the valley).

However, the wind speed and direction vary a lot,

and is dependant on the location of the sampling

site. In June (Fig. 3) the stations located in the

western part of Santiago (Pudahuel, Cerrillos) have

higher wind speeds in the afternoon. Because the

wind comes from the west, it brings clean air into

this part of the city. However, at night the wind that

comes from the mountain (east) does not reach

Pudahuel or Cerrillos and there is very high

ARTICLE IN PRESS

0

20

40

6080

100

120

140

160

Jan

Feb

March

April

May

JuneJu

lyAu

g.Se

pt.Oc

t.No

v.De

c.

PM10(g/cm

3)

La Florida Las Condes

Parque O'higgins Pudahuel

Fig. 2. PM10 monthly average in four stations in Santiago in the

year 2000.


7/28/2019 1-s2.0-S1352231006004377-aula3-mestrado

6/12

atmospheric stability in this part of the city. This

effect can be seen in Fig. 3, the stations located in

the western part of the city (Pudahuel and Cerrillos)

have high wind speeds between 14:00 and 19:00, but

very low speed between 20:00 and 6:00. During the

afternoon the wind comes from the south-west,

which corresponds to a direction between 200 and

2401, as can be seen in Fig. 4. At night the wind

comes from the southeast (1502001). Wind

characteristics during the other winter months are

similar to the observations presented for June.

Throughout summer (DecemberMarch), the

sites located in the western part of the city

(Pudahuel, Cerrillos) have higher PM10 because

the stations are located at the edge of the city, close

to undeveloped land. In this area wind blown dust,

increases the PM10. As seen in Fig. 3, the wind speed

in summer is much higher in Pudahuel and

Cerrillos, up to 6 m s1 than the stations at the

central or eastern part of the city, Parque or Las

Condes, 3.5 m s1. This difference can explain the

higher average PM10 that is measured in Pudahuel

or Cerrillos. It is important to note that the high

PM10 concentration that is seen in Pudahuel during

summer is probably not harmful because most of it

is natural dust: Ca, Al, Si, Ti, Fe and Sr ( Artaxo

et al., 1999). In summer, PM10 levels in Las Condes

and Parque OHiggins are lower than the other

stations (as seen in Fig. 2) because in these sectors of

the city most streets are paved and the wind speed is

less. Thus, re-suspended dust from the west does not

reach Parque OHiggins or Las Condes stations.

ARTICLE IN PRESS

June 2000

0

1

2

0 2 4 6 8 10 12 14 16 18 20 22 24

Speed(m/s)

Pudahuel Cerrillos Parque

La Florida Las Condes

January 2000

0

1

2

3

4

5

6

7

8

0 2 4 6 8 10 12 14 16 18 20 22 24

Speed(m/s)

Pudahuel

Cerrillos

La Florida

Las Condes

2.5

1.5

0.5

Parque

Time Time

Fig. 3. Hourly average wind speed in January (summer) and June (winter) for several stations of the Macam network.

0

Direction(deg)

Cerrillos Parque La FloridaLas Condes Pudahuel

Cerrillos Parque La FloridaLas Condes Pudahuel

0

50

350

0 2 4 6 8 10 12 16 18 20 22 24

Direction(deg)

300

250

200

150

100

TimeTime

140 2 4 6 8 10 12 16 18 20 22 2414

January 2000 June 2000

300

250

200

150

100

50

Fig. 4. Hourly average wind direction in January (summer) and June (winter) for several station of the Macam network.


7/28/2019 1-s2.0-S1352231006004377-aula3-mestrado

7/12

To illustrate the correlation of particle matter

with traffic and wind speed, the PM10 hourly

average for several months has been calculated for

two stations, Pudahuel and Las Condes, respec-

tively (Figs. 5 and 6). PM10 for the Pudahuel station

(Fig. 5) is shown for several months of the year2000. January, March and November correspond to

warmer months, June and July corresponds to the

colder months of the year. This plot shows several

interesting trends in the PM10. For all months it is

possible to discern a peak at 8:00 which is most

likely due to vehicular emissions during the morning

rush hour. In the warmer months (January, March

and November) there is a peak in PM10 occurring at

20:00 h, decreasing at 24:00 h, showing the influence

of the evening rush hour. In the colder months

(June, July) the maximum is shifted towards later

hours, peaking at 23:0024:00 h. This increase iscaused by a reduction of atmospheric turbulence

that also reduces dispersion of pollutants and it is

not related to traffic. As shown in Fig. 3, the wind

speed at night decreases to 11.5 m s1. In addition,

during winter large temperature differences can

occur between day and night (up to 25 1C). Cooling

of the surfaces earth at night generates a tempera-ture inversion that reduces the air turbulence. This

effect leads to a well-known accumulation of

pollution in this area of Santiago (Rutland and

Garreaud, 1995; Gramsch et al., 2000). Results for

the other winter months and for most other stations

are similar showing the same pattern.

The data for Pudahuel indicate that vehicular

emissions have a clear influence on PM10 only in the

morning. In the evening a clear influence from the

rush hour traffic on the PM10 is seen only in summer

(November, January and March). During the other

hours PM10 seems to be influenced by the wind andtemperature inversion.

In the afternoon (12:0018:00 h) there is a

decrease in PM10 in Pudahuel which is due to an

increase in the wind speed and clean air coming

from the west. The wind speed and direction shown

in Figs. 3 and 4, confirm this fact. It has to be noted

that PM10 decreases in the afternoon in spite of the

fact that emissions from vehicles remain approxi-

mately constant (because the flux of vehicles does

not decrease much).

The only station in which PM10 has a differentpattern is Las Condes. Fig. 6, shows the PM10hourly average. The peak in the morning or evening

cannot be seen, indicating that there is little

influence from traffic. Instead, a wide peak in the

afternoon is observed, which is most likely due to

transport of pollution from downtown. The wind

pattern shown in Figs. 3 and 4 indicates that in

Parque OHiggins in the afternoon the wind

direction is 2302501, i.e. directed towards Las

Condes with a speed of 22.5 m s1. This wind can

carry pollution from downtown towards Las Con-

des site. Another feature of the data in Fig. 6 is that

the PM10 peak at night due to the temperature

inversion cannot be seen for any month in Las

Condes station, while in Pudahuel it is very clear in

June and July. The altitude of the Las Condes site is

about 250 m higher than Pudahuel, probably

located near or above the inversion layer. This

could also explain the lower concentrations seen at

this station.

At night and early morning, the wind coming

from the north-east cleans the eastern part of

Santiago (Las Condes), but it takes the pollution

ARTICLE IN PRESS

00 2 4 6 8 10 12 14 16 18 20 22 24

Con

centration(ug/m3)

JanuaryMarchJuneJulyNovember

Time (hr)

Pudahuel

250

200

150

100

50

Fig. 5. PM10 hourly average in Pudahuel station in the year 2000.

0

20

40

60

80

100

120

140

0 2 4 6 8 10 12 14 16 18 20 22 24

Concentration(g/m

3)

JanuaryMarch

JuneJuly

November

Las Condes

Time (hr)

Fig. 6. Hourly average of PM10 at Las Condes station in the year

2000.


7/28/2019 1-s2.0-S1352231006004377-aula3-mestrado

8/12

from downtown towards the west, therefore in-

creasing PM10 levels in Pudahuel, Cerrillos and El

Bosque. As seen in Fig. 3, at night the wind

decreases in magnitude as it reaches the central

and western part of Santiago, thus it is not strong

enough to clean the area.

3.2. Ozone data

Ozone is a secondary pollutant that is generated

through reactions of NO2 and NO with O2 in the

atmosphere with intervention from UV radiation.

Thus, ozone is generated primarily in summer and

during the hours when the UV radiation is a

maximum. In Fig. 7, hourly average of ozone

concentrations are shown for several months in

2000 for the Las Condes site. The correlation

between ozone concentration and UV radiation is

clearly seen because the shape of the UV radiation

curve is very similar to the concentration curve.

Although not shown, results for the other months of

the year show a similar shape. The data for the

stations in the east part of the city (La Florida,

Providencia and La Paz) are also similar, and peak

with UV radiation. However, in the stations located

in the west part of the city (Pudahuel, Cerrillos, El

Bosque), the O3 concentration shape is different

than the UV radiation shape (Fig. 8), remaining

high into the evening. In this area of the city, theaverage ozone levels are lower, and the peak is not

as pronounced because the station is located upwind

from the center of Santiago. Therefore, the NO and

NO2 generated in downtown do not reach these

stations and only local pollutants are responsible for

the generation of ozone.

3.3. Cluster analysis

The geographical distribution of the stations in

the Macam network in Santiago was not the result

of a study of pollution levels, but the stations were

placed in sites thought to be representative of large

sectors of the city. One of the aims of this study is to

determine if there is redundancy of the stations, and

whether there are sectors of the city that have

similar concentration levels and pollution patterns.

The cluster analysis outlined in Section 2.4 was

applied to the PM10 and O3 data for the year 2000.

The results of the process for PM10 are shown inTable 2. When Ri 0:724, the stations are sepa-

rated into two groups, with each group having the

smaller possible intra-variance. The groups corre-

spond to stations located in the west-central part of

the city (Pudahuel, Parque and Cerrillos) or the

eastern part (Providencia, La Paz, La Florida, El

Bosque, Las Condes). If a higher Ri is imposed, i.e.

that the members of the group are more related to

each other, more groups start to appear. Las

Condes station breaks away and forms a separate

group, indicating that the behavior of the station is

different than all of the others. This can also be seen

in Figs. 5 and 6, which show that the temporal

behavior of PM10 in Las Condes is very different

than Pudahuel (or the other stations). There is an

increase in PM10 in Las Condes in the afternoon

(1216 h), and Pudahuel has a decrease in PM10. If

Ri is set to 0.855, four groups of stations are

obtained, in which it is possible to see a topogra-

phical trend. The stations in the central-west part of

the city are grouped together; the stations in the

south (El Bosque and La Florida) and the stations

in the north (Providencia and La Paz) are also

ARTICLE IN PRESS

0

10

20

30

40

50

60

70

80

90

100

0 2 4 6 8 12 16 18 22

Time

O3(ppb)

March

June

August

October

December

Las Condes

10 14 20 24

Fig. 7. Average O3 concentration at Las Condes station in the

year 2000.

0

10

20

30

40

50

60

0 6 8 24

Time

O3(p

pb) March

June

AugustOctoberDecember

Pudahuel

2 4 12 1410 16 18 20 22

Fig. 8. Average O3 concentration at Pudahuel station in the year

2000.


7/28/2019 1-s2.0-S1352231006004377-aula3-mestrado

9/12

grouped. Las Condes station remains isolated. A

diagram of the groups that are formed is shown in

Fig. 9.

The same type of analysis can be carried out with

the O3 data and the results of the calculation are

shown in Table 3. In this case, if Ri 0:

909 twogroups are obtained: one located in the east and

north (La Paz, Providencia, Las Condes), and one

located in the west and south (Pudahuel, Parque,

Cerrillos, El Bosque and La Florida). As with PM10,

the groups are related to the geographical location

of the stations. If we set Ri to a higher value, more

groups start to appear. If Ri 0:956, the same

groups as with PM10 are obtained. Las Condes

station breaks away and forms a separate group, the

stations in the central-west part of the city are

grouped together; the stations in the south (El

Bosque and La Florida) and the stations in thenorth (Providencia and La Paz) are also grouped.

Again, the geographical location of the station

determines how the stations are clustered.

It should be noted that the configuration of the

groups for O3 is the same as for PM10, in spite of the

fact that these pollutants have very different sources

and have maximum concentrations in different

season of the year. O3 is a secondary pollutant

generated during the day, when the UV radiation

has a maximum and PM10 is a primary pollutant

that has many different sources. The fact that bothpollutants have very similar distribution is a strong

indication that the concentration levels are primar-

ily determined by the topographical and meteor-

ological characteristics of the area. These results

also indicate that the pollution generated over the

ARTICLE IN PRESS

Table 2

Results of the cluster analysis of PM10 data

Iteration Cluster

no

Group

members

Intra-

variance

Ratio of intra

to total

variance, Ri

1 1 All 4.990 0.624

2 1 Parque 2.341

Pudahuel

Cerrillos

2 Providencia 3.453

La Paz 0.724

La Florida

El Bosque

Las Condes

3 1 Parque 2.341

Pudahuel

Cerrillos

2 Providencia 3.057

La Paz 0.799La Florida

El Bosque

3 Las Condes 1.000

4 1 Parque 2.341

Pudahuel

Cerrillos

2 La Florida 1.744

El Bosque 0.855

3 Las Condes 1.000

4 Prov. 1.756

La Paz

5 1 Parque 1.696

Cerrillos 0.899

2 La Florida 1.744El Bosque

3 Las Condes 1.000

4 Prov. 1.756

La Paz

5 Pudahuel 1.000

Fig. 9. (a) Clustering of the city into two groups when Ri 0:72 and (b) clustering into four groups when Ri 0:86. The analysis was

performed with PM10 data.


7/28/2019 1-s2.0-S1352231006004377-aula3-mestrado

10/12

city is redistributed in four large areas according to

the meteorological and topographical conditions of

the area. Three of these areas have two or more

monitoring sites. The findings of this study indicate

that in cities like Santiago, the actions required to

reduce pollution have to be directed towards the

whole city. The local sources may have a minor

effect on the local concentration levels.

4. Discussion

Information from Santiagos monitoring network

was used to study the seasonal and geographical

trends of PM10 and O3 in the year 2000. The

pollutant concentrations from two stations are

found to have very different behavior. Pudahuel

station has the highest PM10 levels in winter and it is

located in the lowest part of the city (480 m above

sea level). Las Condes is a station located in the

eastern part of the city, close to the Andes mountain

range, about 220 m higher than Pudahuel. It shows

the lowest average PM10 levels in winter, but in

summer it has the highest ozone levels. Thesedifferences seem to be related to the meteorological

and topographical diversity of the sites.

The PM10 data from the monitoring stations of

the Metropolitan Air Quality Monitoring Network

show a pronounced dependence with the season of

the year. PM10 in summer is 50% lower than winter,

as seen in Fig. 2, and this difference is most likely

due to the higher winds prevalent in summer and the

higher vertical dispersion due to higher tempera-

tures (Gramsch et al., 2000). The wind speed in

summer can be as high up to 6 m s1 compared to

2.5ms1 in winter (Fig. 3). All stations but one,show a peak in PM10 at 8:00 indicating that during

the rush hour there is a strong influence from traffic.

However, this influence is not seen during the rest of

the day. The data in Fig. 2 show that in the

afternoon, there is a decrease in PM10 although the

traffic remains approximately constant. The de-

crease is due to stronger winds in the afternoon

(Fig. 3). The increase in PM10 from traffic in the

evening rush hour (18:0022:00 h) is only seen in

summer (January, March and November in Fig. 5).

In winter, the increase in PM10 occurs between 21:00and 24:00 h, which is partly related to an increase in

traffic, but primarily it is related to a decrease in

wind speed (Fig. 3) and temperature inversion

(Gramsch et al., 2000).

The PM10 pattern in Las Condes station is also

related to meteorological and topographical condi-

tions, because in the afternoon the downtown wind

(Fig. 4) is directed towards Las Condes and can

carry pollution from downtown. A similar effect,

but with opposite direction, occurs with the wind at

night and early morning. In the eastern part of

Santiago (from 20 to 6 h) the wind speed (Fig. 3)

close to the mountains is higher than in the western

side of the city. The wind speed for Las Condes is

clearly higher than the other stations and the

direction (between 70 and 1001) corresponds to

wind coming from the east. The wind brings clean

air from the mountains, reducing the PM10 levels at

night in this part of the city.

Cluster analysis of the data indicates that the

PM10 and O3 pollution generated over the city is

redistributed in four large areas. It is interesting to

note that the grouping depends on the location of

ARTICLE IN PRESS

Table 3

Results of the cluster analysis of O3 data

Iteration Cluster no Group

members

Intra-

variance

Ratio of intra

to total

variance

1 1 All 6.895 0.862

2 1 Parque 4.54

Pudahuel

Cerrillos

La Florida

El Bosque 0.909

2 Prov 2.73

La Paz

Las Condes

3 1 Parque 2.81

Pudahuel

Cerrillos

2 Prov 2.73

La Paz 0.932Las Condes

3 La Florida 1.9

El Bosque

4 1 Parque 2.82

Pudahuel

Cerrillos

2 La Florida 1.907

El Bosque 0.956

3 Las Condes 1.000

4 Prov 1.93

La Paz

5 1 Parque 1.00

2 La Florida 1.91

El Bosque3 Las Condes 1.000

4 Prov 1.93 0.967

La Paz

5 Pudahuel 1.92

Cerrillos


7/28/2019 1-s2.0-S1352231006004377-aula3-mestrado

11/12

the stations, which probably is due to the topo-

graphical characteristics of the site where the station

is located. Pudahuel is the sector of the city with the

lowest altitude (450 m above sea level), with lower

temperatures and higher humidity in winter. There

is a hill towards the north (Renca hill) that mayprevent good ventilation. The south part of the city

(El Bosque and La Florida) is very flat, with no hills

nearby. Providencia and La Paz are located close to

several hills, in a sector of Santiago slightly higher

than the rest of the city. Three of these areas have

two or more monitoring sites. The results indicate

that in cities like Santiago, reduction of pollution

has to be directed towards the whole city because

local pollution levels are not solely determined by

local sources. For example, pollution from kerosene

and wood burning used for house heating may drift

to the lowest part of the city (Pudahuel) generatingthe large PM10 levels observed in winter.

There are other cities with complex topographical

and meteorological conditions, like Beijing, Bogota,

Mexico City or Athens (Molina et al., 2004) in

which the distribution of pollution is influenced by

the topography of the site. In these situations, the

methods and results of this study may be used to

suggest pollution-control guidelines.

5. Conclusions

The results show a pronounced dependence of the

concentration levels with the season of the year,

with PM10 being higher in winter and O3 in summer.

In winter, the PM10 maximum occurs during the

night, which is an indication that the meteorological

conditions are responsible for the high levels. The

higher sector of the city does not show the PM10increase at night, suggesting that the height of the

temperature inversion occurs at lower altitude.

Cluster analysis of the data indicates that PM10and O

3

generated over the city is redistributed in

four large areas. The areas are the same for O3 and

PM10, in spite of the fact that these pollutants have

very different sources and have their maximums on

different season of the year. The fact that both

pollutants have similar distribution is a strong

indication that the concentration levels are primar-

ily determined by the topographical and meteor-

ological characteristics of the area. These results

indicate that in cities like Santiago, reduction of

pollution has to be directed towards the whole city

because local pollution levels are not solely deter-

mined by local sources.

Acknowledgements

This study was supported by the National

Commission for the Environment (CONAMA)

under contract no 20127600-2. We thank Laura

Jeria for the statistical calculations.

References

Artaxo, P., Oyola, P., Martinez, R., 1999. Aerosol composition

and source apportionment in Santiago de Chile. Nuclear

Instruments & Methods in Physics Research Section B

Beam Interactions with Materials and Atoms 150 (14),

409416.

Cifuentes, L.A., Vega, J., Kopfer, K., Lava, L.B., 2000. Effect of

the fine fraction of particulate matter versus the coarse mass

and other pollutants on daily mortality in Santiago, Chile.

Journal of the Air and Waste Management Association 50 (8),

12871298.Conama, 1994. Environmental Base Law. March 9, 1994, Chilean

Government, Santiago, Chile. http://www.conama.cl/rm/568/

article-931.html.

Conama, 1997. Metropolitan Region Prevention and Deconta-

mination Plan. July 1, 1997, Edited by Comisio n Nacional

del Medio Ambiente, Conama RM, Santiago, Chile. http://

www.conama.cl/rm/568/article-932.html.

Dockery, C.A., Pope III, C.A., Xu, X., Spengler, D.J., Ware,

J.H., Fay, M.E., Ferris, B.G., Speitzer, F.E., 1997. An

association between air pollution and mortality in six U.S.

cities. New England Journal of Medicine 329, 17533744.

Estraus, A description of the program can be found in http://

www.sectra.cl/contenido/modelos/transporte_urbano/Estraus_

ingles.htm.

Gramsch, E., Catala n, L., Ormen o, I., Palma, G., 2000. Traffic

and seasonal dependence of the light absorption coefficient in

Santiago de Chile. Applied Optics 39 (27), 48954901.

Ilabaca, M., Olaeta, I., Campos, E., Villaire, J., Tellez-Rojo,

M.M., Romieu, I., 1999. Association between levels of fine

particulate and emergency visits for pneumonia and other

respiratory illnesses among children in Santiago, Chile.

Journal of the Air and Waste Management Association 49,

154163.

Jorquera, H., Perez, R., Cipriano, A., Espejo, A., Letelier, M.V.,

Acuna, G., 1998. Forecasting daily maximum levels at

Santiago, Chile. Atmospheric Environment 32 (20),

34153424.Lee, S.A., Hastie, T., Mancilla, P.F., Astudillo, P.O., Kuschner,

W.G., 2000. Fine particulate air pollution (PM2.5) and

medical visits for lower respiratory tract illnesses among

children in Santiago, Chile. Journal of Investigative Medicine

48 (1), 524.

Molina, L.T., Molina, M.J., Slott, R.S., Kolb, C.E., Gbor, P.K.,

Meng, F., Singh, R.B., Galvez, O., Sloan, J.J., Anderson,

W.P., Tang, X., Hu, M., Xie, S., Shao, M., Zhu, T., Zhang,

Y.H., Gurjar, B.R., Artaxo, P.E., Oyola, P., Gramsch, E.,

Hidalgo, D., Gertler, A.W., 2004. Air quality in selected

megacities, critical review online version. Journal of the Air &

Waste Management Association 55.

Ostro, B.D., Eskeland, G.S., Sanchez, J.M., Feyzioglu, T., 1999.

Air pollution and health effects: A study of medical visits

ARTICLE IN PRESS

http://www.conama.cl/rm/568/article-931.htmlhttp://www.conama.cl/rm/568/article-931.htmlhttp://www.conama.cl/rm/568/article-932.htmlhttp://www.conama.cl/rm/568/article-932.htmlhttp://www.sectra.cl/contenido/modelos/transporte_urbano/Estraus_ingles.htmhttp://www.sectra.cl/contenido/modelos/transporte_urbano/Estraus_ingles.htmhttp://www.sectra.cl/contenido/modelos/transporte_urbano/Estraus_ingles.htmhttp://www.sectra.cl/contenido/modelos/transporte_urbano/Estraus_ingles.htmhttp://www.sectra.cl/contenido/modelos/transporte_urbano/Estraus_ingles.htmhttp://www.sectra.cl/contenido/modelos/transporte_urbano/Estraus_ingles.htmhttp://www.conama.cl/rm/568/article-932.htmlhttp://www.conama.cl/rm/568/article-932.htmlhttp://www.conama.cl/rm/568/article-931.htmlhttp://www.conama.cl/rm/568/article-931.html

7/28/2019 1-s2.0-S1352231006004377-aula3-mestrado

12/12

among children in Santiago, Chile. Environmental Health

Perspectives 107 (1), 6973.

Perez, P., Trier, A., Reyes, J., 2000. Prediction of PM2.5

concentrations several hours in advance using neural net-

works in Santiago, Chile. Atmospheric Environment 34 (8),

11961198.

Rutland, J., Garreaud, R., 1995. Meteorological air-pollutionpotential for Santiago, Chiletowards an objective episode

forecasting. Environmental Monitoring and Assesment 34 (3),

223244.

Sanhueza, P., Vargas, C., Jimenez, J., 1999. Daily mortality in

Santiago and its relationship with air pollution. Revista

Medica de Chile 127 (2), 235242.

SAS Institute Inc., http://www.sas.com.

Schmitz, R., 2004. Modeling of air pollution dispersion in

Santiago de Chile. Atmospheric Environment, in print.

Silva, C., Quiroz, A., 2003. Optimization of the atmosphericpollution monitoring network at Santiago de Chile. Atmo-

spheric Environment 37 (17), 23372345.

ARTICLE IN PRESS

http://www.sas.com/http://www.sas.com/

1-s2.0-s1352231006004377-aula3-mestrado

Documents